AI search answers can use affiliate reviews and comparison pages as source material because those pages often organize information around buyer questions. The engine’s answer is a synthesis, not an independent product test. A citation does not show that an affiliate commission caused the page to be selected, that the page is reliable, or that the answer accurately represents everything it says.
How an AI search answer gets from a question to a recommendation
In broad terms, an AI search system can retrieve web pages relevant to a query, use material from them to produce a response, and show links to sources. The system’s wording is generated; it is not necessarily a direct quotation or a complete summary of any one cited page. How retrieval and answer generation work varies by platform, so one company’s description should not be treated as a universal account of every AI search engine.
What Google says about its own AI search features
Google says its AI search features draw on relevant, up-to-date pages in Google’s Search index. Its Search Central generative AI guide describes retrieval-augmented generation, or grounding, and query fan-out: for a broad question, the system may issue related searches to find material addressing different parts of it. It can then generate an answer and provide links to pages that support it. That is Google’s explanation of Google features, not proof that ChatGPT, Gemini, Perplexity, or other systems use the same architecture.
For Google AI Overviews and AI Mode, a page must be indexed and eligible for a Search snippet to be eligible as a supporting link. Google also says that meeting its requirements does not guarantee crawling, indexing, or appearing in an answer. Eligibility is a threshold, not a promise of citation.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →#1 Best Overall
Why an affiliate review can fit a product question
A query such as “best robot vacuum,” “best life insurance,” “best X compared,” or “which X should I choose” asks for a decision, not just a product definition. A comparison page may put relevant material in one place: a shortlist, selection criteria, specifications, prices, strengths and weaknesses, and reasons for recommending one option over another. That structure can make it useful source material for an answer assembled around a buyer’s question.
That usefulness does not establish that affiliate status caused an engine to retrieve or cite the page. Nor does a citation mean the engine endorses the publisher, recommends its methods, or has confirmed that the publisher tested each product. The page may be one input among several, and its presence in an answer does not tell a reader how much weight the system gave it.
What the recent citation figures do—and do not—show
A 2026 Suff Digital industry analysis offers a broad snapshot of affiliate publishers appearing in product-answer citations. It collected 17,524 answers to 2,000 “best product or service” questions across 20 shopping categories and five AI products on September 23, 2026. Its classification counted roughly 150 publishers as affiliate publishers based on earning affiliate commissions.
| Reported result | What was counted | How to interpret it |
|---|---|---|
| 70% of answers cited at least one affiliate review or comparison publisher | Answers in Suff Digital’s specific question panel and five-product sample | An observed rate for that sample and publisher classification, not a universal share of AI answers or proof that commissions influenced citation selection. |
| 33.9% of citations in the average answer that cited any source were from affiliate review sites | Citations within answers containing sources | This is a share of visible citations in sourced answers. It is not a measure of every page an engine retrieved or considered internally. |
The findings indicate that affiliate review and comparison publishers were a substantial visible citation source in this study. They do not show that all affiliate pages are cited often, that a reader will click a citation, or that the same proportions hold for another prompt set, date, geography, or platform. The analysis is observational: it measures what appeared, not why the system chose it.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Why a citation is not proof that the answer is right
A citation gives readers a route to inspect a source; it does not guarantee that every sentence beside it is supported by that page. A 2026 preprint study by Haofei Xu, Umar Iqbal, and Jacob M. Montgomery examined Google AI Overviews using 55,393 queries over 40 days, from March 13 to April 21, 2026. Of 98,020 decomposed claims, 11.0% were unsupported by the cited pages. In that study, the measured problem included claims that lacked support in the page cited for them. This is preprint evidence about that study’s queries and method, not a general error rate for every AI engine or answer.
For a consequential purchase, open the cited page and check whether it supports the specific claim being made. Then check primary product documentation or other direct evidence for details such as compatibility, specifications, exclusions, and current prices. A page may be relevant to the topic without substantiating a particular claim in the generated answer. Ofcom’s work on answer engines also highlights the need to consider source traceability and how answers represent their sources.
Affiliate incentives and what responsible disclosure looks like
An affiliate publisher may earn a commission when a reader follows a tracked link and completes a qualifying action. That financial relationship is relevant context for judging a recommendation; its presence does not by itself prove that a review is poor, just as an affiliate link does not prove that a product was independently tested.
The Federal Trade Commission’s Endorsement Guides FAQ says, “The closer the disclosure is to your recommendation, the better.” It advises that disclosures be clear and conspicuous so readers can assess the weight to give an endorsement. The phrase “affiliate link” alone may not explain that the publisher receives payment. A publisher should also accurately say whether products were independently tested, provided for free, or selected by another method; it should not imply hands-on testing that did not happen.
Google’s search-link guidance addresses a separate technical issue: “We ask sites participating in affiliate programs to qualify these links with rel="sponsored", regardless of whether these links were created manually or dynamically.” That HTML attribute is for link qualification; it does not replace a plain-language disclosure to readers.
Affiliate monetization is not a quality verdict. The ACCC’s 2024 inquiry report summarizes concerns raised in earlier work about highly optimized, low-quality affiliate pages ranking highly in organic search. That is a documented concern about some pages, not a description of every affiliate review. For a recommendation to be assessable, readers need to see the review method, selection criteria, relevant limitations, and evidence behind the verdict.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What AI answers may mean for publisher visits and affiliate revenue
A visible answer can change whether a person visits a publisher, but available studies do not establish one universal revenue effect for affiliate sites. They measure different populations and outcomes, and their results should not be collapsed into a single prediction.
- Transaction data: A 2026 study in Marketing Science observed more than 50,000 ChatGPT-origin transactions across 973 ecommerce websites over August 2024–July 2025, alongside 164 million transactions from other channels. The authors caution that last-click attribution can understate channels that helped with earlier discovery. This is evidence about transactions and attribution in the study, not a forecast of affiliate commissions.
- Click-through experiment: In a preregistered 2026 field experiment with 1,100 participants, removing AI Overviews and AI Mode increased publisher click-through in the experiment’s setting. An AI Mode-only experience reduced click-through as well as reported user experience and trust. Those experimental results concern the tested conditions; they are not a measurement of every publisher’s traffic or affiliate earnings.
Together, these findings show why referral patterns matter, while leaving the effect on any particular affiliate publisher unsettled. A user may see an answer without clicking, click a source after using the answer to narrow choices, or arrive through a channel not captured by last-click reporting.
Quick Recap
What publishers should prioritize—and what readers should check
For publishers
- Make the decision criteria explicit and explain why each option fits them; do not rely on a bare ranking.
- Provide original, useful evidence where possible, with current specifications and prices clearly dated and sourced. Google’s Search Central guidance says, “A first-hand review provides a unique perspective based on personal experience, whereas a summary of existing content simply restates information already available elsewhere.” This is publisher guidance, not a guarantee that firsthand reviews will rank or be cited.
- Describe what was tested, how products were chosen, and what the review cannot establish. Distinguish direct testing from research based on manufacturer information or other sources.
- Put a clear commission disclosure close to recommendations and affiliate links, and qualify affiliate links with
rel="sponsored"as Google requests. - Follow foundational search guidance and build useful, distinctive pages rather than chasing supposed AI-only tricks. Google says third-party services cannot guarantee performance; Search Console is its first-party tool for measuring a site’s Search performance.
For readers
- Treat an AI recommendation as a starting point, not an independent product test.
- Open the cited page and confirm it actually supports the claim that matters to your choice.
- Check the publisher’s disclosure and review method, then verify important details against direct product information.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
